What is with these articles that put layoff statistics next to discussions of AI, not directly making the claim that they are connected (because it is clearly baseless), but clearly leaving it for the reader to make the connection. I find this to be incredibly dishonest.
>Just to point out, this particular article does make the claim these are connected.
yeah, only for one paragraph:
> Last year, companies including Chegg, IBM and Dropbox cited the onset of AI as a reason to rethink staffing. More recently, Duolingo and even Google have suggested the same as they seek to mobilize resources to capitalize on the AI boom.
And everything before and after that is either fluff or does not make the claim that they're connected. For instance layoffs at Amazon was brought up early in the article, but it's unclear how that has to do with AI. Google is brought up as well, but if you click through the source, the layoffs has less to do with "AI replacing workers" (as the article suggests) and more "google is laying off in the hardware part of the business because they want to focus on the AI part".
> layoffs has less to do with "AI replacing workers" (as the article suggests)
The article is pretty clear that it isn't just talking about AI directly replacing workers. From the second sentence of the article: "...not all the recent layoffs in the tech industry are directly linked to AI tools simply replacing workers" and goes on to explain from there that they are talking about companies changing what they are spending money on.
Well, the dollars are related? R&D spending increasing in one area but decreasing in another. Companies have finite dollars to spend on R&D so to some extent this is zero-sum.
How is it dishonest to present where major companies are cost-cutting and where they are investing, then highlighting the obvious connection across these companies?
> How is it dishonest to present where major companies are cost-cutting and where they are investing, then highlighting the obvious connection across these companies?
The article doesn't really try to argue "the obvious connection across these companies" aside from a couple of sentences. Most of the article is spent discussing off layoffs and AI investments without attempting to link the two.
Is it realkly dishonesty? The fearmongering around AI and capitalism is also not irrational so it would require some sort of goal to qualify as lying. What we are seeing is a complete lack of understang the technology and to be honest, it is extremly difficult to dissect and even harder to know whats behind the tech layoffs.
The hype train has no brakes. This kind of stuff will continue until some of these companies have such steep declines in revenue due to crappy products that they file for bankruptcy. I just seriously hope this doesn’t lead to another AI winter, because the progress that has been made is quite real, quite good, and waiting for someone to find that killer use case to transform human lives for the better.
it kind of cant. the last ai winter was based on the fact that you needed expensive experts and got really marginal results. now that you can do something useful with even commodity machines its just not going to go away.
training workloads that cost tens of millions of dollars might
This has nothing to do with AI. AI can't replace a decent newgrad. It literally can't find a React import and cannot be relied on for anything without a human double checking it, which takes as long as writing it yourself, when it comes to line-of-business applications.
The only people espousing AI magic are those who used it for "lol generate me an HTML/JavaScript book about AI that has pagination" or "lol build me a linked list" or serious researchers in traditional AI areas.
This has everything to do with COVID over-hiring and a ton of people coasting because they are used to doing nothing from when everyone was desperate for devs.
> AI can't replace a decent newgrad. It literally can't find a React import and cannot be relied on for anything without a human double checking it, which takes as long as writing it yourself, when it comes to line-of-business applications.
This is little-picture, short-sighted thinking. A one-for-one replacement of a specific kind of employee is not a requirement to statistically displace employees across a workforce. AI tooling can augment the productivity of existing employees, which results in a lessened need for human labour. And you're making the baseless assumption that we have achieved the pinnacle of AI development. This problem will only get worse as AI tooling improves.
Let me be more specific - as an architect working on a senior dev task today, it's faster for me to type out the code / figure things out then prompt ChatGPT to do the problem. I actually have it pulled up all day in vain attempts to use it. So far I had it remove access modifiers and {get; set;} from a property list so I could copy it into a UML diagram. In reality, it would have been faster if I just remembered the shortcut.
I would say this is archaic, luddite and unambitious thinking. Did the need for labor decrease after the lightbulb, electricity, or the telegraph? Or did Quality of life go up while projects expanded in scope and we accomplished greater things with more efficiency?
> Did the need for labor decrease after the lightbulb, electricity, or the telegraph? Or did Quality of life go up while projects expanded in scope and we accomplished greater things with more efficiency?
It absolutely did...and you literally said it yourself. What is "more efficiency" except a reduced need for labour for a fixed unit of "productivity"?
Indeed. This is the hangover from the era of cheap money and private equity gone wild. It’s a healthy adjustment for the industry that puts great people back in circulation, to be picked up by healthier companies that didn’t blow their brains out on cheap capital. It also frees people up to start new companies because suddenly just working for FAANG doesn’t make you rich anymore.
When I looked for people a year ago, the market was still tight and frankly things seemed hopeless. We struggled along with an undersized team. I now have really good people walking up unsolicited. That hasn’t happened for many years. It’s a welcome change.
I have a slightly different opinion. I agree that we humans have a creative spark that drives research, true art, etc. However, I am 90% seeing AI tools as a means to let me create more, and faster. I have co-pilot integrated with my Emacs and VSCode setups, and it is hard to imagine turning back time and not having co-pilot available. Yesterday I wanted a snappy title for a new open source project (inspired by last Friday’s All-In podcast) so I wrote up a three paragraph description and asked mixtral-8-7B (running locally on my Mac) as well as Anthropic’s Claude 2 and ChatGPT for project name suggestions. Lots of good ideas, I chose my favorite.
Last February I was writing a book on ‘safe AI’ and after I wrote each chapter, I had ChatGPT read it and asked for additional topics I may have neglected. It was also good for generating end of chapter summaries that I re-worded to fit my style.
For the last ten years I have tried to be on a ‘digital diet’ - limiting time on digital devices except for high productivity sprints. This includes limiting time on HN and X. It feels good to use AI tools to get more done quickly so I have more time to cook elaborate meals for my family, hike and hang out with my friends, and read.
Yes, during the pandemic staffing levels surged so even these 15% cuts aren’t even getting close to pre-pandemic numbers. People also forget that even though a company may be classified as Tech, they hire a lot of people in different professions than software to keep the business going - sales, accounting, HR, marketing, etc. Tech is also only like 2% of jobs in the US so layoffs in the tech sector, especially advertising tech (meta, google), aren’t indicative of cross-industry layoffs.
I could see after some time that custom AI apps/prompts could supplant traditional search that has been around for 25 years. SEO will become irrelevant when custom AI tools exist that narrow the focus and allow follow-up questions and context.
I wouldn't be surprised if is partly due to AI, but not because of the tech itself, but because of the additional compensation for talent and infrastructure spending it currently brings.
Not tech, but AI is absolutely replacing people, just not wholly. To take an example, a friend of mine is releasing a small book of D&D one-shots she wrote. She was prepared to pay someone to make some art, but found that between AI and some editing skills, she got it close enough that she doesn't need to hire anyone. It's small scale of course, but that's money no longer going towards artists and instead to AI companies.
Newflash: people don’t just need some minimum wage money to be happy.
UBI might be better than nothing, but boy are people going to be lost without social mobility and purpose. Hopefully we can find that some other way if all labor is ever replaced. Which I am skeptical about.
Or we could have a generation of amazing musicians, programmers, painters and social workers.
Have you read David Graeber's Bullshit Jobs? He talks about how a lot of the British music scene of the 60's and 70's was basically funded by social welfare programs. Broke kids having time to hone their craft.
But the "AI" of today can just copy and style, can make art faster than people.
I think that while we can say, art is art, people just like doing art, which I think is true for some people, there was also a major inventive to make good music in the 60s and 70s which will no longer be there if a machine can write an album in 20 seconds and steal your style almost instantaneously.
Maybe people will be using AI to make unimaginably good art and prosper from that, which I hope so but as it stands now, it's a very different situation to what happened in the past.
> He talks about how a lot of the British music scene of the 60's and 70's was basically funded by social welfare programs. Broke kids having time to hone their craft.
Just because somenthing happened organically/by chance in the past, doesn't automatically mean you can just replicate the same result today by forcing thse same policies from the 60's in the world of today. You might definitely get a completely different outcome.
We had something like that during the pandemic. Kids, or young adults, were blowing it all their easy money on crypto and $GME stonks which led to tonnes of get-rich-quick scammers and pyramid schemes, not making new music or art at home.
Also, the generation of The Baatles didn't have their brains fried by endless doom scroling on tiktok, social media and Fortnite. If they were bored at home with no job, they had little escapism so listening to other artists and creating their own competing content was the go-to, while today if you've got no job and are sitting bored at home there's million other hedonistic avenues of keeping yourself entertained that are much easier than having to pick up a guitar, learn to play it well and come up with meaningful lyrics.
Passionate artists and musicians will always find a way to create content they enjoy even if they'll still be broke as every generation before them. The Beatles of today are on twitch or youtube, creating content about the stuff their passionate about and getting funded via Patreon or other donations.
I have another anecdote to bring to the table. I'm from Eastern Europe and the best music, especailly hip-hop, that came out of my post-communist hellhole country was in the 90's when we had the worst possible economy, rampant inflation and crime, no jobs, no welfare, no money, no opportunities, nothing. But those times produced the best artists and football players(soccer for you guys). Now the economy and welfare is 100x better than back then, but the quality of art and music has gone down the toilet. Why is that? I think the new found prosperty brought with it a career focused culture and because now there's more and better things to do for money and entertainment than struggling with creating good music in a garage while broke.
So to me, it seems the most depressive economic periods produce the best art and music, and the abundant "easy times" produce the worst and least inspiring. It was similar to 50 Cent's "Get rich or die trying" or Eminem's "8 Mile", all these artist came from suffering of economic harship, not economic prosperity, but nobody sane in the head wants to bring back the hard times just so we could get better hip-hop and music.
NFTs were the new music and art, once you cared to look past your given Bored Apes. I also don't think that it's a coincidence that XR and AI hit their strides during that period. There are more examples.
And, note that the momentum-killer in all these cases was not organic, but institutions clawing back their riches - by hook, and certainly by crook, in many cases (I'm glad you mentioned GME). Return-to-office had every rational reason not to happen, and they forced it anyway.
Career focused culture has destroyed a large amount of social cohesion. If machines do more work, humans can go back to community building. Type A personalities that demand social mobility can play politics to see who leads their respective tribe. The rest of us can spend the time learning about the world, developing engaging hobbies, reading philosophy and experiencing great works of art. Maybe just focus on a personal spirituality and/or self-actualization.
There is so much humans should be doing to make our brief existence a little more happy and fulfilling. Little of this has to do about contributing labor.
I do think this sounds like a plausible outcome. I can even see the "AI wealthy" getting a bit lonely at the top and wanting to be part of the human race again.
Notice how rich people wear hoodies and train at the gym like everyone else now?
>Career focused culture has destroyed a large amount of social cohesion.
Eaxtly, spot on!
> If machines do more work, humans can go back to community building.
In an ideal utopic society, yes, but that's not what's happening. More automation replacing workers just means more people out of work from older jobs, means more competition and lower wages for the remaining few un-automatable jobs flooded with now jobless workers.
It's impossible to pull off if all the profits from automation go to those doing the automation instead of the disposed workers, but if those in charge of investing and reasearching automation would not get huge profits they wouldn't be motivated to invest in automation in the first place and we'd be back to the middle ages.
So what do we do? What's the realistic sollution? Do we go back to communism and have the state in charge of all economic planning and development? Maybe not since that doesn't works too well in the long run.
Ideally there should be a middle ground that's missing right now where replacing workers with autoamtion is done through union negociations at a steady pace to enure disposed labor can be retrained in newer careers.
The value of UBI is it means your ability to survive isn't tied to having a job, giving you a much stronger negotiating position. It's the same as socialized health care, housing, subsidized or socialized food...if you have to risk your life to find a better offer, it's harder to walk away from the table.
In theory, base on supply and demand, jobs should be priced at
difficulty * value * unpleasantness
Instead, a lot of the hardest, most important, and most unpleasant jobs pay some of the lowest wages, because they default land on the people with the least negotiating power (undocumented immigrants, prisoners, or those so stuck in crushing poverty they have to take the first option they get), and the employer gets to get away with paying well under the "market" value of the labor. So many jobs wouldn't exist at their current price point (if at all), if there was a basic minimum amount of "You can just about get by" money that everyone had.
There are many basic social needs that aren't being met for a majority of the population because so much time and energy has to go towards merely surviving. I think people who have a vested interest in this paradigm overestimate how much the average person "needs" work for "social mobility and purpose", versus, say, companionship, community, education, curiosity, housing stability, etc. Maybe they're afraid that people will realize that the rat race and daily grind don't fill the hole, but stuff that doesn't cost money or that is difficult to profit from do.
>There are many basic social needs that aren't being met for a majority of the population because so much time and energy has to go towards merely surviving.
Always has been since humanity's ~200k year history. The unpriviledged masses having free time for leisure and entertainment between work and sleep that isn't a Sunday, is a relatively new development of about ~100 years or less, for which much blood has been spilled. My grandparents were doing subsistance farming working about ~14h/day, 6-7 days/week just to survive, while today I can have my needs met by clocking in 8h of work in an office 5 days/week.
Ideally we should strive for better, but we forget our ancestors had it way worse not too long ago.
>Always has been since humanity's ~200k year history. The unpriviledged masses having free time for leisure and entertainment between work and sleep that isn't a Sunday, is a relatively new development of about ~100 years or less
Incorrect. If anything, the period during and immediately preceding the Industrial Revolution (before those hard-won labor victories) is the outlier. The common thread in the traditional societies that still exist is a great deal of leisure time or light physical labor, even for farmers. Through the Middle Ages and up to the the advent of chattel slavery and industrialization in the Americas and Europe, labor schedules were quite mild and left plenty of time for leisure, socializing, labor outside of primary production duties (chores, cooking, etc., through which creative faculties could be exercises); much of the literature describing back-breaking labor, written by the privileged classes, was actually a form of propaganda or wish-fulfillment fiction.
I'd suggest reading The Overworked American by Juliet Schor. It provides an overview of historical modes of labor in the West.
This is what I'm thinking everytime I see union comments in layoff posts. Either those unions will be very weak because of AI or AI will be artificially weaken by them, which is not the best case either.
As a current AI researchers, my opinion is while AI may surpass the average human soon (if not already) in a game of 9s humans are always better. So I wouldn’t trust a system with 0 humans in it for failsafe.
The transition will be painful either way - mass unemployment, aggravated by a steep deflationary cycle, mass boredom and hopelessness in a redundant workforce conditioned from birth to find purpose in paid work, and the cultural and political upheaval from the permanent consolidation of power and wealth by the tiny elite supplying and developing the AI who are suddenly bigger than any or all governments.
Do most people find purpose in paid work? I know some young adults about 25 years old who want to retire early (seems to be a movement). I personally would value being a provider and career but work is just a means to that goal and not the end-goal itself.
Recently it indeed seemed like there are more layoffs than usual. But then I went to layoffs.fyi (no affiliation) and at least according to their data, which seems reliable, if anything we see a decline in layoffs. Jan and Feb 2023 were pretty bad so we'll need to see in a month how this year compares.
Ten years ago a bartender told he was studying economics and asked me what field I was in. I said I study math and computer science. He says “all that stuff is going to be automated” and I replied “I wonder what that will do to our economy”. All the blood drained out of the poor guy lol
